The accident at the Fukushima Daiichi nuclear power plant (NPP) in Japan has put again the spotlight on the issue of hydrogen explosion and its consequences. About that, so called lumped-parameter (LP) codes remain, for the time being, the customary tool for simulating hydrogen combustion in actual NPPs. Namely, the volumes of actual containments are much too large for performing calculations with CFD codes.

Thus, estimations of hydrogen combustion consequences still rely on LP codes, which use relatively simple models of combustion.Although hydrogen combustion in single compartments has already been simulated quite often, there were not many opportunities so far to simulate flame propagation between compartments. With this purpose, a benchmark exercise was organized, based on a laboratory-scale hydrogen combustion experiment, performed in a parallelepiped shape double-compartment vessel of 1.48 m3 total volume at the University of Pisa (Italy).

In the experiment, many hydrogen combustion tests were performed by varying the initial hydrogen concentration and the area of the orifice in the wall separating the two compartments.Three different tests were simulated and the main results, which consist of time dependent pressure and temperature, are presented, compared between them and to experimental results (for the pressure increase). This work is focused on the assessment of flame propagation models of the codes, with the aim of identifying their limitations as regards to that complex phenomenon.